Golden-free Hardware Trojan Detection

This direction is merged into another direction: hardware intellectual property theft. Its last version stays readable and citable here.

Version 1 of 118 papers33 conceptsfrozen 2026-10-04

In a globalized semiconductor supply chain, designers often integrate third-party Intellectual Property (IP) cores into their systems. This creates a critical security gap: how do you verify that a circuit is trustworthy if you don't have a "golden model"—a guaranteed clean version of the design—to compare it against? Traditional verification relies on finding discrepancies between a suspect chip and a reference, but in modern manufacturing, that reference often doesn't exist.

Recent research has shifted toward golden-free detection, a paradigm that identifies malicious logic (Hardware Trojans) by analyzing the inherent structural and statistical properties of a circuit netlist rather than looking for differences from a baseline.

The Geometry of Malicious Logic

The most effective way to understand a circuit's intent is to treat it as a graph. In this view, logic gates are nodes and wires are edges. Researchers like Lilas Alrahis and Ozgur Sinanoglu argue that integrated circuits are naturally topological. While a Trojan designer can try to hide malicious gates among thousands of legitimate ones, they cannot easily hide the unique "shape" of a trigger mechanism.

Hardware Trojans typically consist of a trigger (which waits for a rare condition) and a payload (which executes the attack). This structure creates anomalous topological patterns. By using Graph Convolutional Networks (GCNs), frameworks like GATE-Net, developed by Nikhil Muralidhar and Ryan Gerdes, can learn to recognize these suspicious structural motifs directly from raw netlists. Instead of needing a golden model, the model learns the "language" of circuit connectivity to spot logic that looks out of place.

Detecting Stealth Through Information Entropy

If structure provides the "where," signal activity provides the "what." Hardware Trojans are designed to be stealthy, meaning they rarely transition or "flip" during normal operation to avoid detection by power analysis.

Renjie Lu and Huawei Li proposed HTDet, which uses information entropy to identify these dormant threats. By analyzing the transition probability of signals, they can cluster logic gates based on their activity levels. Malicious triggers appear as statistical outliers—regions of abnormally low entropy that deviate from the functional logic of the chip. This unsupervised approach allows for detection without any labeled training data or reference designs.

From Comparison to Characterization

The transition to golden-free verification represents a move from comparative analysis to behavioral characterization.

  1. Gate-Level Characterization: Li Zhang and Chip-Hong Chang demonstrated that one can establish a "power fingerprint" of a design through simulation. This creates a virtual baseline for process variation, allowing physical chips to be audited without needing a physical golden twin.
  2. Statistical Anomaly Detection: Approaches like PL-HTD, developed by Chen Dong and Wenzhong Guo, use Principal Component Analysis (PCA) to treat Trojans as simple statistical noise. By focusing on the characteristics of "normal" circuitry, any logic that doesn't fit the mathematical profile of the rest of the chip is flagged.

By combining graph-based structural awareness with unsupervised statistical modeling, these techniques allow engineers to secure the supply chain even when the "ground truth" of a design is unavailable.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_golden_free_hardware_trojan_detection_a3e5322a,
  title = {Golden-free Hardware Trojan Detection},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/golden-free-hardware-trojan-detection/rdr_a3e5322a68404a3f91b15682afa6f864},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_golden_free_hardware_trojan_detection_a3e5322a_v1,
  title = {Golden-free Hardware Trojan Detection},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/golden-free-hardware-trojan-detection/rdr_a3e5322a68404a3f91b15682afa6f864/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}